EXPERIMENTAL EVOLUTION 149 
187. Methods. The best of all experiments in evolution are those that 
are constantly being made in nature. Such experiments are readily dis- 
covered and studied in the case of origin by adaptation; variants present 
much greater difficulties, while mutants are very rare under natural condi- 
tions. The method which makes use of these experiments may be termed 
the method of natural experiment. The number of ecads which appear 
naturally in vegetation is limited, however, and it is consequently very 
desirable to produce them artificially, by the method of habitat culture. 
This method, while involving more labor than the preceding, yields results 
that are equally conclusive, and permits the study of practically every 
species. The method of control culture, which is carried on in the plant- 
house, naturally does not possess the fundamental value of the field methods. 
Tt is an invaluable aid to the latter, however, since it permits the physical 
factors to be readily modified and controlled. All these methods are based 
on the indispensable use of instruments for the measurement of physical 
factors. 
METHOD OF NATURAL EXPERIMENT 
188. Selection of species. Species that are producing variants or ecads 
are found everywhere in nature; those which give rise to mutants seem, 
however, to be extremely rare. Consequently, mutants can not be counted 
upon for experimental work, and their study scarcely needs to be con- 
sidered. When a mutant is discovered by some fortunate chance, the 
mutable species from which it has sprung, and related species as well, 
should be subjected to the most critical surveillance, in the hope that new 
mutants will occur or the original one reappear. On account of the sud- 
denness with which they appear, mutants do not lend themselves readily to 
natural experiment, and after they have once been discovered, inquiry into 
the causes and course of mutation is practicable only by means of habitat 
and control cultures. Among variable species, those are most promising 
that show a wide range of variation and are found in abundance over 
extensive areas. A species which occurs in widely separated, or more or 
less isolated areas, furnishes especially favorable material for investigation, 
since distance or physical barriers partly eliminate the leveling due to con- 
stant cross-fertilization. The individuals or groups which show appreciable 
departure from the type are marked and observed critically from year to 
year. The direction of the variation and the rapidity with which small 
changes are accumulated can best be determined by biometrical methods. 
Representative individuals of the species and each of its variants should 
likewise be selected from year to year. After being photographed, these are 
preserved as exsiccati, and with the photographs constitute a complete 
